The initial sample included 2,103 youths; the follow-up sample had 1,018 participants. Results at follow-up revealed gender differences. Males were more drug-involved than were females. However, ethnicity was unrelated to drug involvement at either point in the study. The general distributions of drug use remained reasonably stable across groups, but considerable change into and out of high drug involvement occurred. Consistent with peer cluster theory, this change was predicted by peer drug use and peer requests of the participants to use drugs and to a lesser extent by the individual's willingness to ask others to avoid using drugs. Findings suggested the importance of peer group processes on the naturalistic development and reduction of high drug involvement. Findings also indicated the importance of studying changing peer processes to improve understanding of change in drug involvement over time. Table and 32 references (Author abstract modified)